U.S. patent application number 10/999066 was filed with the patent office on 2005-06-02 for apparatus and method for inspecting patterns on wafers.
Invention is credited to Ahn, Byung-Seol, Cho, Jae-Sun, Kim, Joo-Woo, Lee, Byung-Am, Lee, Chang-Hoon, Lee, Sung-Man.
Application Number | 20050119844 10/999066 |
Document ID | / |
Family ID | 34617409 |
Filed Date | 2005-06-02 |
United States Patent
Application |
20050119844 |
Kind Code |
A1 |
Lee, Chang-Hoon ; et
al. |
June 2, 2005 |
Apparatus and method for inspecting patterns on wafers
Abstract
A wafer pattern inspecting apparatus and method are disclosed.
The apparatus comprises an image sensor to acquire image data from
a reference die and a sample die, an external memory to store the
image data, an encoder to compress the data, a decoder to
decompress the data, an internal memory device to store the
compressed image data of the reference die, an arithmetic module to
process the image data for the reference dies to extract a
reference image data, a reference storage memory to store
compressed reference image data, and a comparison module to compare
the sample die image data with the reference image data to an
extract defect data for the sample die.
Inventors: |
Lee, Chang-Hoon;
(Cheonan-si, KR) ; Lee, Byung-Am; (Suwon-si,
KR) ; Ahn, Byung-Seol; (Suwon-si, KR) ; Cho,
Jae-Sun; (Suwon-si, KR) ; Kim, Joo-Woo;
(Seoul, KR) ; Lee, Sung-Man; (Suwon-si,
KR) |
Correspondence
Address: |
VOLENTINE FRANCOS, & WHITT PLLC
ONE FREEDOM SQUARE
11951 FREEDOM DRIVE SUITE 1260
RESTON
VA
20190
US
|
Family ID: |
34617409 |
Appl. No.: |
10/999066 |
Filed: |
November 30, 2004 |
Current U.S.
Class: |
702/83 |
Current CPC
Class: |
G01N 21/95607
20130101 |
Class at
Publication: |
702/083 |
International
Class: |
G06F 015/00 |
Foreign Application Data
Date |
Code |
Application Number |
Dec 1, 2003 |
KR |
2003-86368 |
Claims
What is claimed is:
1. An apparatus for inspecting die patterns on a wafer, comprising:
an image sensor to acquire image data from a reference die and a
sample die; an external memory to store the image data; an encoder
to compress the image data; a decoder to decompress a compressed
image data; an internal memory to store compressed image data
associated with the reference die; an arithmetic module to process
the image data associated with the reference dies to extract a
reference image data; a reference storage to store compressed
reference image data; and a comparison module to compare the image
data associated with the sample die with the reference image data
to extract pattern defect data for the sample die.
2. The apparatus of claim 1, wherein each one of the encoder,
decoder, arithmetic module, and comparison module divide image data
into a two-dimensional block, and process the image data in
relation to the two-dimensional block.
3. The apparatus of claim 1, wherein the image sensor converts the
image of the sample die to image data defined by a two-dimensional
array of pixels.
4. The apparatus of claim 1, wherein the internal memory stores
compressed image data for a plurality of reference dies.
5. The apparatus of claim 1, wherein the arithmetic module
processes a block data having a same coordinate value selected from
the image data of the reference dies, each being divided into the
two-dimensional block to extract reference image data from the
respective block.
6. The apparatus of claim 1, wherein the comparison module compares
image data for the sample die divided into the two-dimensional
block with the block data having the same coordinate value selected
from the reference image data to extract defect data for the sample
die.
7. The apparatus of claim 6, further comprising a result storage
memory to store defect data, and an output module for outputting
the defect data.
8. The apparatus of claim 7, wherein the result storage memory
stores defect data compressed by the encoder.
9. The apparatus as recited in claim 7, wherein the output module
directly outputs the defect data, or outputs a result stored in the
result storage memory.
10. The apparatus as recited in claim 7, wherein the result storage
memory stores defect data compressed by the encoder, and the output
module outputs the defect data stored in the result storage memory,
wherein the defect data is restored at the decoder to be
transmitted by the output module.
11. The apparatus as recited in claim 7, wherein the result storage
memory stores a wafer defect map composed of the defect data.
12. The apparatus as recited in claim 11, wherein the output module
outputs the wafer defect map stored in the result storage
memory.
13. A method for inspecting die patterns formed on a wafer,
comprising: sampling a plurality of reference dies; acquiring image
data from the plurality of reference dies, and compressing the
acquired image data; dividing the respective compressed image data
into a plurality of blocks, and processing the compressed image
data block by block to extract reference image data; acquiring
image data from a sample die; and comparing the sample die image
data with the reference image data to extract defect data for the
sample die.
14. The method of claim 13, wherein the sample die is selected from
the same wafer as the plurality of reference dies.
15. The method of claim 13, wherein the image data is composed of a
plurality of two-dimensionally pixel arrays, and wherein the image
data further comprises position data and corresponding image
data.
16. The method of claim 13, wherein the image data and the
reference image data are each divided into a plurality of blocks
and compared block by block to extract the defect data.
17. The method of claim 13, wherein compressing the acquired image
data further comprises: acquiring image data for a selected
reference die; and dividing the image data for the selected
reference die into a plurality of blocks and compressing the image
data for the selected reference die block by block.
18. The method of claim 13, wherein the extracting the reference
image data further comprises: dividing the respective compressed
image data for the reference dies into "n" blocks; selecting an
n.sup.th block of the respective compressed image data; restoring
the selected n.sup.th block; processing the restored n.sup.th block
to extract reference data from the n.sup.th block; and
reconfiguring the reference data for the n.sup.th block to define
the reference image data.
19. The method as recited in claim 18, wherein each of the blocks
comprises a two-dimensionally array of "m" pixels, wherein
processing the n.sup.th block comprises: selecting an m.sup.th
pixel from each of the respective n.sup.th blocks; comparing image
data for each of m.sup.th pixel from each of the respective
n.sup.th blocks; selecting a median value for the n selected
m.sup.th pixels to be defined a m.sup.th reference pixel; and
defining the reference data for the n blocks in relation to the
selected median values.
20. The method of claim 13, wherein the extracting the defect data
comprises: dividing the reference image data and the sample die
image data into "n" blocks; selecting an nth block from the
reference image data and the sample die image data; comparing the
selected nth blocks to extract defect data for the n.sup.th block;
and, extracting the defect data from the "n" blocks to configure
defect data for the sample die.
21. The method of claim 20, wherein the reference image data and
the sample dies image data are composed of two-dimensionally pixel
arrays comprising image data and position data, wherein each of the
n blocks is composed of two-dimensional array of "m" pixels, and
wherein extracting the defect data from the n blocks comprises:
selecting the mth pixel from a respective n.sup.th block of the
reference image data and the sample die image data; comparing the
image information of the m.sup.th pixels to determine a defect; and
extracting the defect data for the n.sup.th block in relation to
the compared m.sup.th pixels.
22. The method of claim 21, wherein extracting the defect
comprises: setting a threshold value for pixel image data for the
reference image data; and comparing sample die image data to the
threshold value.
Description
BACKGROUND OF THE INVENTION
[0001] 1. Field of the Invention
[0002] The present invention generally relates to an apparatus and
a method for inspecting patterns formed on a wafer. More
particularly, the present invention generally relates to an
apparatus and a method for inspecting dies on a semiconductor wafer
to determine the position of pattern defects.
[0003] A claim of priority is made to Korean Patent Application No.
2003-86368, filed on Dec. 1, 2003 in the Korean Intellectual
Property Office, the disclosure of which is incorporated herein in
its entirety by reference.
[0004] 2. Description of the Related Art
[0005] A conventional semiconductor wafer includes a plurality of
dies where integrated circuit devices are printed thereon. The
integrated circuit device is fabricated by complex processes,
including a series of inspection steps performed on patterns formed
on the integrated circuit device. A complete inspection is
typically performed to verify the accuracy of the patterns formed
on the wafer. One conventional inspection method is a die-to-die
method. The die-to-die method is a method where a sample die is
compared with its adjacent dies.
[0006] The conventional die-to-die method will now be described
with reference to FIG. 1.
[0007] As shown in FIG. 1, a conventional wafer has a plurality of
two-dimensionally partitioned dies. In the die-to-die method, a
sample die 1 is inspected by comparing the patterns formed on die 1
with the patterns formed on adjacent dies 2 and 3.
[0008] In the case where a die 4 is disposed at an edge of a wafer,
adjacent dies 5 and die 6 are selected as reference dies, because
there is only one immediate adjacent die 5. Here, die 4 is a
non-functional die, because it is an incomplete die. If a pattern
at a specific position on sample die 4 is identical to at least one
of reference dies 5, 6, that position is determined as defect-free,
i.e., normal.
[0009] In another scenario, if a similar defect occurs at the same
position on both reference dies, the sample die will be deemed
defective even if that position on the sample die is normal.
Furthermore, in the case where the sample die and one of the
reference dies are similarly defective at the same position, the
pattern corresponding to this position on the sample die will be
deemed normal. As described above, die 4 disposed at the wafer edge
is non-functional. When die 5 is inspected, because die 4 is
incomplete and non-functional, and if die 6 is defective, die 5
will be deemed as defective even though it is normal.
[0010] The die-to-die method is inaccurate; therefore, a better
inspection method is required.
SUMMARY OF THE INVENTION
[0011] In one aspect, the present invention provides an apparatus
and a method for inspecting patterns on a die and accurately
determining pattern defects on the die.
[0012] In another aspect, the present invention provides an
apparatus and method for accurately inspecting patterns on a sample
die without reference to adjacent dies.
[0013] In still another aspect, the present invention provides a
pattern inspecting apparatus for extracting a reference value to
accurately determine a defect in a sample die, and a pattern
inspecting method for inspecting defects in sample dies using the
reference value.
[0014] In order to achieve these aspects, the present invention
provides an apparatus for inspecting die patterns on a wafer having
an image sensor to acquire image data from a reference die and a
sample die, an external memory to store the image data acquired by
the image sensor, an encoder to compress the image data, a decoder
to decompress a compressed image data, an internal memory device to
store compressed image data for the reference die, an arithmetic
module to process the reference die image data to extract a
reference image data, a reference storage to store the compressed
reference image data, and a comparison module to compare the image
data from the sample die with the reference image data to extract
pattern defect data for the sample die.
[0015] The present invention also provides a method for inspecting
die patterns formed on a wafer by sampling a plurality of reference
dies, acquiring image data from the plurality of reference dies,
compressing the acquired image data, dividing the respective
compressed image data for each one of the plurality of reference
dies into a plurality of blocks, and processing the compressed
image data for the respective blocks to extract reference data,
acquiring image data from a sample die, and comparing the image
data from the sample die with the reference image data to extract
defect data for the sample die.
BRIEF DESCRIPTION OF THE DRAWINGS
[0016] FIG. 1 is a top plan view of a wafer to explain the
conventional die-to-die method.
[0017] FIG. 2 is a block diagram of a pattern inspecting apparatus
according to a preferred embodiment of the present invention.
[0018] FIG. 3 is a flowchart for explaining a pattern inspecting
method according to a preferred embodiment of the present
invention.
[0019] FIG. 4 is a flowchart showing steps of setting a reference
image data in the pattern inspecting method according to the
preferred embodiment of the present invention.
[0020] FIG. 5 is a flowchart showing steps of inspecting a sample
die in the pattern inspecting method according to the preferred
embodiment of the present invention.
[0021] FIG. 6 through FIG. 8 shows compression of image data of a
sample die according to the preferred embodiment of the present
invention.
[0022] FIG. 9 and FIG. 10 show a method for extracting reference
image data from compressed image data of the sample die according
to the preferred embodiment of the present invention.
[0023] FIG. 11 shows steps of inspecting a sample die according to
a preferred embodiment of the present invention.
DETAILED DESCRIPTION OF THE INVENTION
[0024] The present invention will now be described more fully with
reference to the accompanying drawings, in which preferred
embodiments of the invention are shown. The invention may, however,
be embodied in different forms and should not be construed as
limited to the embodiments set forth herein. Rather, these
embodiments are provided as teaching examples of the invention.
Like numbers refer to like elements throughout the
specification.
[0025] FIG. 2 is a block diagram of a pattern inspecting apparatus
according to a preferred embodiment of the present invention.
Referring to FIG. 2, the pattern inspecting apparatus includes: an
image sensor 50 to acquire image information from a die on a wafer;
an external memory 52 to store the acquired image information
(data); an encoder 54 to compress the image data; a decoder 58 to
decompress the compressed image data; a reference storage 62 to
store reference image data; and, a comparison module 64 to compare
sample die image data with the reference image data. The pattern
inspecting apparatus further comprises an internal memory 56 and an
arithmetic module 60 activated during a procedure to extract
reference image information from a plurality of sample dies.
Arithmetic module 60 and comparison module 64 are preferably
embedded in the same data processing unit. The pattern inspecting
apparatus may further comprise an output module 68 to output
detected defect data, and a result storage memory 66 to store the
detected defect data.
[0026] In further detail, image sensor 50 acquires image data from
a die, typically including an array of two-dimensional pixels. Each
of the pixels corresponds to position data on the die and to image
data for that position. The smaller the size of the pixel, the more
accurately a pattern can be inspected. For example, as pixel size
decreases in proportion to a minimum pitch size for the sample
integrated circuit device; the ability to accurately acquire an
image data is inversely proportional to the pixel size. Thus, as
pixel size decreases, the resulting quantity of the image data
increases. The volume of the image data also increases with
segmentation of measurement values for the image data as well as
the pixel size. As the measurement values of the respective pixels
are segmented, the ability to measure micro defects is enhanced. In
order to accurately detect micro defects, it is preferable to
reduce the pixel size and to segment the measurement values for the
image data, even where the resulting volume of image data increases
accordingly.
[0027] Image sensor 50 sequentially acquires image data in the form
of a two-dimensional array of pixels. The acquired data is
transmitted and stored in external memory 52. External memory 52
acts as a buffer to temporarily store the acquired image data.
Preferably, external memory 52 has sufficient storage capacity to
store all image data acquired from all sampled dies. However, with
the recent trend toward smaller patterns, it may be difficult to
process the entire volume of image data acquired from a plurality
of dies. Therefore, in the present invention, the acquired data is
optionally compressed in order to reduce the volume of data
requiring storage and processing.
[0028] Image sensor 50 preferably acquires image data from a
reference die and a sample die. A reference image data setting
procedure is preferably divided into an image data memory procedure
for the reference die and a reference image data memory procedure.
In the image data memory procedure related to the reference dies,
image sensor 50 acquires image data from at least one sample
reference die. If a plurality of sample reference dies are selected
according to their positions on the wafer, image sensor 50 acquires
the image data from the sample reference dies and transmits the
acquired data to external memory 52. The image data for one or more
sample reference dies is preferably acquired and transmitted
depending on the volume of image data for the die and the capacity
of external memory 52. The sample reference die image data stored
in external memory 52 is divided into a plurality of blocks and
compressed for each respective block, and then eventually
transmitted to internal memory 56. Each of the blocks includes data
related to the two-dimensionally array of pixels and corresponding
block position data. The sample reference die image data for the
respective block is transmitted to encoder 54. Encoder 54
compresses the image data and transmits the compressed block data
to internal memory 56. If the capacity of external memory 52 is
insufficient, the block data transmitted to encoder 54 may be
erased from external memory 52 after a period of time. If the block
data transmitted to internal memory 56 is reconfigured according to
the corresponding position data to transmit a final compressed
block data. In this manner, compressed image data for the sample
reference die is completely stored. By repeating the above
procedure, image data for the sample reference die is compressed
and stored in internal memory 56.
[0029] After storing image data for the reference dies is complete,
the reference image data storing procedure preferably begins.
However, the reference image data storing procedure may
alternatively start at the time when first block data of compressed
image data from the final sample reference die is transmitted to
internal memory 56. In the reference image data storing procedure,
image data from the sample reference dies stored in internal memory
56 are restored for each block to be processed by arithmetic module
60. The block data having the same corresponding position data as
the respective sample reference die is selected for transmission to
decoder 58. Arithmetic module 60 performs a series of processing
procedures on the restored block data to extract reference block
data. The reference block data is compressed at encoder 54 and
sequentially stored in reference storage 62. When the reference
data for the last block is transmitted to reference storage 62, the
reconfiguration of the reference image data is complete.
[0030] In the procedure to inspect a sample die, i.e., a
to-be-inspected die, image sensor 50 acquires image data for the
sample die. The acquired image data is transmitted to external
memory 52. Comparison module 64 compares the image data for the
sample die with the reference image data at the respective block in
order to detect a defect. The respective block reference image data
stored in reference storage 62 is restored, and then transmitted to
comparison module 64. Also, the respective block image data for the
sample die, which is stored in external memory 52, is transmitted
to comparison module 64. In this case, the blocks to be compared
have the same position data. The defect data detected by comparison
module 64 is transmitted to output module 68, and then output or
stored in result storage memory 66. The defect data may be
compressed at encoder 54, and then stored in result storage memory
66. Result storage memory 66 preferably stores the defect data for
each wafer and thereby forms a wafer defect map.
[0031] A pattern inspecting method according to a preferred
embodiment of the present invention will now be described with
reference to the flowchart shown in FIG. 3.
[0032] Referring to FIG. 3, the pattern inspecting method includes
reference image data setting steps 400 and die inspecting steps
500. Steps 400 begin with a sample die selecting step 100 and ends
with an encoded reference image data storing step 112. Steps 500
begin with a sample wafer selecting step including n dies 116 and
ends with defect mapping step 124. The step of selecting a sample
wafer 116 and step of decoding reference image data 114 are
independently performed.
[0033] Steps 400 will now be described hereinafter. On a wafer, "m"
reference sample dies are selected (100). An identification (i) is
assigned to each of the sample dies. A two-dimensional image data
for the first sample is input (102). The two-dimensional image data
is preferably acquired by image sensor 50, and stored in external
memory 52. The two-dimensional image data is encoded (104). The
encoded image data is stored (106). Steps 102-106 are carried out
for all of the reference sample dies. The ID of current reference
sample die is checked to determine whether the data input for all
of the reference samples is complete. When the data input for all
reference samples is stored, a reference image data is extracted
from the samples (108). The extracted reference image data is
encoded (110). The encoded reference image data is stored (112),
and the reference image data setting steps (400) are complete. The
larger the number of reference sample dies, the better the
reference image data. However since the number of reference samples
are proportional to the execution time for the image data setting
steps, the number of reference samples are preferably selected
within a permissible value.
[0034] The die inspecting steps 500 begin when a sample wafer is
selected (116). The sample wafer may be identical to the reference
sample wafer in selected step 100. Independently of steps 500, the
encoded reference image data is decoded (114). The sample wafer
preferably has "n" sample dies. Identification (j) is assigned to
each of the sample dies. Image data forth first die sample is input
(118). The input image data for the first sample die is compared
with the reference image data (120). Defects are determined based
on the comparison between the first die sample and the reference
image data (122). Based on the comparison, defect data of the first
die sample is extracted (122). The ID of a sample die is checked,
and steps 118-122 are repeated until the ID equals the vlaue n.
When defect data for the n sample dies is extracted, a defect map
for the sample wafer is constructed (124).
[0035] FIG. 4 is a flowchart further illustrating the reference
image data setting steps according to a preferred embodiment of the
present invention. First, "m" reference sample dies are selected
(200). The number of selected reference sample dies is preferably
selected in consideration of the accuracy and processing time for
the resulting reference data. For example, three reference sample
dies are selected. An identification (i) is assigned to each of the
reference sample dies. A first reference sample is selected (202).
Each of the reference sample dies is preferably divided into "1"
blocks. An identification (k) is assigned to each of the blocks. A
first block data is encoded in a block unit (204), and then the
encoded block data are stored (206). Each of the blocks includes
corresponding position data for the reference sample die. When the
encoded data for an entire block of the first reference sample die
is stored, the ID of the reference sample die is checked, and steps
202-208 are performed on the next sample. After the completion of
step 208, the ID of the reference sample die is checked. If the
checked ID equals the value m, the block ID (k) is initialized and
the first block data for each reference sample is decoded (210).
The first block data for the reference sample dies, i.e., m first
block data sets are compared (212), and reference data for a first
block is extracted from the first block data (214). All of the
blocks are sequentially decoded, and the decoded blocks are
compared (210-214). When the reference data for the respective
blocks has been extracted, the ID of the block is checked. If the
ID equals the value "1", the extraction of reference data is
complete. The extracted reference position data is arranged to
configure the reference image data (216).
[0036] The die inspecting steps according to the preferred
embodiment of the present invention will now be described with
reference to the flowchart shown in FIG. 5. First, a sample wafer
is selected (300). The wafer has "n" sample dies. An identification
(j) is assigned to each of the sample dies. Two-dimensional image
data for the first die is input (302). A die is divided into "1"
block. An identification (k) is assigned to each of the 1 blocks.
Reference data for a first block and the first block data for the
sample die are retrieved (304 and 306). The reference data is
compared with the sample die block data to determine any defects in
the first block (308). The first defect data is stored (310), and
the block ID is checked. Steps 304-310 are repeated until the block
ID equals the value "1." The defect data for the first sample die
is configured from the defect data extracted when the block ID is
"1" (312). The defect data is preferably output by an output
module. To acquire the defect data for the entire wafer, this
routine proceeds to the next die and steps 302-312 are repeated
until the ID of the current die equal the value "n". When the ID of
the die reaches "n", the extraction of defect data from the n
sample dies is complete. According to the entire die position, a
defect data is used to make a defect map for the wafer (314).
[0037] FIG. 6 through FIG. 8 illustrate compression of image data
for a sample die. Referring to FIG. 6, compression of the image
data is performed by an encoder. Image data for a sample die is
composed of two-dimensional array of pixels and a two-dimensional
image data for the sample die. Image data 10 for a first sample die
is divided into axb block (A.sub.ab). Each block has corresponding
position data for each image. Each block is composed of xxy pixels
(a.sub.xy) each having image data and position data. Block data
temporarily stored in an external memory is sequentially selected
to be transmitted to the encoder. The encoder compresses the
selected block data in order to encode it. The compressed image
data is restored by the lossless image compression method, where
the higher the number of repeated patterns on a die, the better the
compressibility of the data. As a result, compressed block data
A.sub.ab' is reduced in size and composed of changed pixels
a.sub.xy'. The compressed block data A.sub.ab' has position data
for each pixel and position data for the block. Thus, compressed
image data 12 for the sample die is acquired by reconfiguring the
compressed block data A.sub.ab'.
[0038] Referring to FIG. 7, after compressing the image data for
the first sample die is complete, compression of an image data for
a second sample die begins. The image data for the second sample
die is composed of axb block (B.sub.ab), and each of the block is
composed of xxy pixels (a.sub.xy). Similar to the compression of
the first image data, compressed block data B.sub.ab' having
changed pixel data is acquired by compressing the data for the
respective blocks. The compressed block data B.sub.ab' is
reconfigured to acquire a compressed image data 16 for the second
sample die.
[0039] Referring to FIG. 8, compressed image data C.sub.ab' for a
third sample die is acquired in the same manner as described above.
Image data 18 for the third sample die is composed of axb block
(C.sub.ab), each having xxy pixels (c.sub.ab). A compressed image
data 20 is composed of block C.sub.ab' having changed pixels
c.sub.ab'.
[0040] The number of samples is determined in consideration of the
processing speed and accuracy of the reference image data. The size
of a sample die, the size of a block, and the size of a pixel are
preferably identical throughout all the sample dies. Each pixel and
block have corresponding position data. The compressed image data
has the original position data, and its capacity is reduced. Thus
the compressed image data is decoded to restore the original image
data, and decoding is conducted by a lossless compression method to
restore the original image data.
[0041] A method of extracting reference image data from compressed
image data for a sample die will be described with reference to
FIG. 9 and FIG. 10. Referring to FIG. 9, a reference image data is
extracted for each respective block. The compressed block data
A.sub.ab', B.sub.ab', and C.sub.ab' are selected from the
compressed image data 12, 1 6, 20 for the respective sample dies.
The selected block data A.sub.ab', B.sub.ab', and C.sub.ab' have
the same position data. The compressed block data A.sub.ab',
B.sub.ab', and C.sub.ab' are restored. The decoder may decode the
compressed block data A.sub.ab', B.sub.ab', and C.sub.ab'. The
restored block data A.sub.ab, B.sub.ab, and C.sub.ab are processed
to extract reference block data G.sub.ab. Specifically, pixels
a.sub.xy, b.sub.xy, and c.sub.xy having the same position are
selected one by one from the respective block data A.sub.ab,
B.sub.ab, and C.sub.ab, and image data corresponding to the pixel
is processed to define a reference value. In this case, a median of
the pixel image data is defined as a reference value. For example,
in the case where the image data is related to an intensity of
light and its values are 50, 75, and 90 respectively, the reference
value is 75. All pixels are sequentially compared to acquire block
data G.sub.ab having reference values g.sub.xy corresponding to
pixels g.sub.xy. This procedure may be performed by the arithmetic
module.
[0042] Referring to FIG. 10, in the arithmetic module, reference
block data G.sub.ab is transmitted to a reference storage memory.
Specifically, reference block data G.sub.ab acquired at each
respective block is compressed into a compressed block data
G.sub.ab'. Then compressed block data G.sub.ab' is stored in the
reference storage memory. The block data is memorized according its
position data. Simultaneously with the storage of the block data,
reference image data 22 may be configured.
[0043] FIG. 11 shows the steps of inspecting a sample die according
to a preferred embodiment of the present invention. Referring to
FIG. 11, reference image data 22 extracted from the same wafer as
the sample die is selected. Reference image data 22 may be
extracted in advance. An image data 24 of the sample die is
acquired. Block data G.sub.ab is selected block-by-block from
reference image data 22 and image data 24. The reference image data
G.sub.ab is stored while being compressed. Therefore, block data
G.sub.ab' selected from reference image data 22 is restored at a
decoder. A reference pixel (value) g.sub.xy for restored block data
G.sub.ab and a block data T.sub.ab of the sample die are compared
with each other one at a time. The one pixels are compared by a
comparison module that offers a threshold value to the image data
of reference pixel g.sub.xy to determine whether it is defect-free
or not. When the image data of a sample pixel t.sub.xy is within
the threshold value, it is determined to be defect-free. When the
image data of the sample pixel t.sub.xy is below or beyond the
threshold value, it is determined to be defective. This procedure
is sequentially performed for all pixels to configure a defect
block data F.sub.ab. Thus each pixel f.sub.xy of defect block data
F.sub.ab has defect information, not image information. The entire
block of image data 24 of the sample die are compared with
reference image data 22 to acquire defect data 26 for the sample
die. This result is output by an output module. In this case, the
output results may be distinguished by a user. A defect position
and a defect aspect for the sample die may be inspected at the same
time by overlapping image data 24 and defect data 26. The defect
data is stored in a result storage memory. Alternatively, the
defect data is compressed at a decoder for storage or is stored as
a defect map for the wafer by reconfiguring the sample dies.
[0044] As disclosed, reference image data is extracted from a
plurality of selected reference sample dies. By comparing the
reference image data with a sample, defects in the sample die can
be precisely determined. Further when a wafer having successive
defects is inspected, errors caused by adjacent dies can be
prevented. Since image data for a sample die is compressed, the
amount of data processing can be reduced. In addition, the image
data is processed by respective blocks to extract reference image
data and defect data. Therefore, it is possible to achieve
real-time inspection, storage, and transmission.
[0045] While the present invention has been described in detail
with reference to certain preferred embodiments, it should be
apparent that modifications and adaptations to those embodiments
might occur to a person skilled in the art without departing from
the scope of the present invention.
* * * * *